.cling_history shouldn't grow infinitely

[Axel-Naumann]

Explain what you would like to see improved

.cling_history currently grows without limits, we have some that are 144M.

Optional: share how it could be improved

We probably want both:

1. drop the oldest entries, such that it remains < 1M or so.
2. prevent the cling test suite from adding to the history - that's useless.

[dimitry-ishenko]

Maybe follow what bash does with .bash_history? There are two env vars -- HISTSIZE and HISTFILESIZE -- which control how many commands are stored in the file and max file size.

By default HISTSIZE is 500 and HISTFILESIZE is 0.

cling can have two variables -- CLING_HISTSIZE and CLING_HISTFILESIZE -- that are initialized to the same values (500 and 0) by default and allow overriding them.

[dimitry-ishenko]

As regards to the 2nd point, I remember seeing some commits in @vgvassilev's repo that allowed one to specify custom history file. So, the test suite shoud set it to some file using eg mktemp and not pollute the default one.

[dimitry-ishenko]

It looks like point 2 has already been implemented in 90a7639. Just need to get the test suite to specify a temp history file.

[dimitry-ishenko]

And it seem CLING_HISTSIZE has also been implemented in 392d536. It just needs to default to a sane value.

[dimitry-ishenko]

@Axel-Naumann are you still interested in this issue? If so, let me know and I can take a stab at it. Otherwise, I'd hate to do something that will then rot in a neglected PR forevermore.

[Axel-Naumann]

I'm totally interested - I just have very little time slices for this. I'd be amazing if you proposed something! Setting a default, non-infinity value as the upper limit seems the way to go?

[richen666]

@Axel-Naumann @dimitry-ishenko
Is there any progress in this issue? If not, I want to try it.

[dimitry-ishenko]

@richen666 sorry, no progress from me, I've been busy with other things.